Detection method for dangerous behaviors of underground coal mine personnel
-
摘要:
井下人员危险行为检测是煤矿安全防控的关键环节。现有目标检测技术用于人员危险行为检测时，受煤矿井下复杂工况、设备遮挡、多目标密集、粉尘干扰等因素影响，存在特征提取不准确等问题，且未明确界定人员危险行为。以YOLOv8−pose模型为基准架构，采用DCNv4和PConv模块融合的DCNv4−PConv混合模块代替标准卷积，添加混合局部通道注意力（MLCA）模块，并采用感受野注意力卷积（RFAConv）模块替换检测头，构建了PMR−YOLO模型，用于检测井下监控图像中人体关键点，提升检测精度和运算速度。在此基础上设计了人员行为识别算法，将井下人员行为划分为9种类别，基于YOLOv8−pose模型检测的人体关键点形成人体骨架，判断人员行为类别型。采用DsLMF+数据集进行消融实验、对比实验和人员行为识别实验，结果表明：DCNv4−PConv混合模块、MLCA模块、RFAConv模块的引入有效提高了YOLOv8−pose模型的精确度、召回率和平均精度均值（mAP）；PMR−YOLO模型对人体关键点特征提取的精确度、召回率和mAP分别为0.893， 0.841， 0.852，较YOLOv8−pose模型分别提高了6.9%，14.4%，10.5%；基于PMR−YOLO模型的检测方法可有效识别井下人员9种行为类别，识别准确率均不低于96%。
-
关键词:
- 视频识别 /
- 危险行为检测 /
- 人员行为识别 /
- YOLOv8−pose模型 /
- 人体关键点检测
Abstract:Detection of dangerous behaviors of personnel in underground mines is a critical aspect of coal mine safety and risk prevention. Existing object detection technologies face challenges when applied to underground personnel behavior detection, as complex working conditions, equipment occlusion, dense targets, and dust interference often lead to inaccurate feature extraction and undefined behavior classifications. To address these issues, a PMR-YOLO model was constructed based on the YOLOv8-pose architecture. The standard convolution was replaced with a hybrid DCNv4-PConv module, which combines the DCNv4 network and the deformable capability of the PConv module. In addition, a Mixed Local Channel Attention (MLCA) module was integrated into the structure, and the detection head was replaced with a Receptive-Field Attention Convolution (RFAConv) module. These modifications aimed to improve the accuracy and speed of human keypoint detection in underground surveillance images. On this basis, a personnel behavior recognition algorithm was designed to classify underground behaviors into nine categories. Human skeletal structures were generated from the keypoints detected by the YOLOv8-pose model, and behavior types were identified accordingly. The DsLMF+ dataset was used to conduct ablation experiments, comparative experiments, and behavior recognition experiments. The results showed that incorporating the DCNv4-PConv hybrid module, the MLCA module, and the RFAConv module significantly improved the precision, recall, and Mean Average Precision (mAP) of the YOLOv8-pose model. The PMR-YOLO model achieved a precision of 0.893, a recall of 0.841, and an mAP of 0.852 in keypoint detection, representing improvements of 6.9%, 14.4%, and 10.5%, respectively, over the original YOLOv8-pose model. The detection method based on the PMR-YOLO model can effectively identify nine types of underground personnel behaviors, and all recognition accuracies exceed 96%.
-
-
![]()
图 7 数据集中各行为类别样本统计
Figure 7. Sample distribution of each behavior category in the dataset
表 1 消融实验结果
Table 1 Results of model ablation experiments on the MK dataset
模型 精确度 召回率 mAP@0.5 YOLOv8−pose 0.824 0.697 0.747 YOLOv8−pose+DCNv4−PConv 0.885 0.835 0.837 YOLOv8−pose+MLCA 0.881 0.831 0.832 YOLOv8−pose+RFAConv 0.884 0.834 0.836 YOLOv8−pose+DCNv4−PConv+MLCA 0.887 0.836 0.845 YOLOv8−pose+DCNv4−PConv+RFAConv 0.889 0.839 0.843 YOLOv8−pose+MLCA+RFAConv 0.888 0.835 0.841 PMR−YOLO 0.893 0.841 0.852 
下载: 导出CSV
表 2 对比实验结果
Table 2 Comparison experimental results of different models
模型 精确度 召回率 mAP@0.5 推理时间/ms YOLOv8−pose 0.824 0.697 0.747 20 YOLOv8−W6−pose 0.856 0.797 0.829 24 YOLOv8s−pose 0.873 0.813 0.824 19 MoveNet 0.851 0.802 0.815 24 YOLOv11−pose 0.836 0.783 0.812 22 文献[12] 0.864 0.806 0.832 22 PMR−YOLO 0.893 0.841 0.852 17
下载: 导出CSV
表 3 人员行为识别准确率
Table 3 Recognition accuracy of personnel behaviors
% 行为类别 准确率 行为类别 准确率 摔倒 97.2 蹲 97.5 翻越 96.8 走路 98.3 坐 96.5 站立 98.6 倚靠 97.4 工作 97.2 弯腰 98.4
下载: 导出CSV
-
[1] 王国法,王虹,任怀伟,等. 智慧煤矿2025情景目标和发展路径[J]. 煤炭学报,2018,43(2):295-305. WANG Guofa,WANG Hong,REN Huaiwei,et al. 2025 scenarios and development path of intelligent coal mine[J]. Journal of China Coal Society,2018,43(2):295-305.
[2] 王佟,刘峰,赵欣,等. “双碳”背景下我国煤炭资源保障能力与勘查方向的思考[J]. 煤炭科学技术,2023,51(12):1-8. DOI: 10.12438/cst.2023-0095 WANG Tong,LIU Feng,ZHAO Xin,et al. Reflection on China's coal resource guarantee capacity and exploration work under the background of "double carbon"[J]. Coal Science and Technology,2023,51(12):1-8. DOI: 10.12438/cst.2023-0095
[3] 王国法,张建中,薛国华,等. 煤矿回采工作面智能地质保障技术进展与思考[J]. 煤田地质与勘探,2023,51(2):12-26. DOI: 10.12363/issn.1001-1986.23.02.0062 WANG Guofa,ZHANG Jianzhong,XUE Guohua,et al. Progress and reflection of intelligent geological guarantee technology in coal mining face[J]. Coal Geology & Exploration,2023,51(2):12-26. DOI: 10.12363/issn.1001-1986.23.02.0062
[4] 王海军,曹云,王洪磊. 煤矿智能化关键技术研究与实践[J]. 煤田地质与勘探,2023,51(1):44-54. DOI: 10.12363/issn.1001-1986.22.12.0992 WANG Haijun,CAO Yun,WANG Honglei. Research and practice on key technologies for intelligentization of coal mine[J]. Coal Geology & Exploration,2023,51(1):44-54. DOI: 10.12363/issn.1001-1986.22.12.0992
[5] 马宏伟,王世斌,毛清华,等. 煤矿巷道智能掘进关键共性技术[J]. 煤炭学报,2021,46(1):310-320. MA Hongwei,WANG Shibin,MAO Qinghua,et al. Key common technology of intelligent heading in coal mine roadway[J]. Journal of China Coal Society,2021,46(1):310-320.
[6] 张旭辉,杨文娟,薛旭升,等. 煤矿远程智能掘进面临的挑战与研究进展[J]. 煤炭学报,2022,47(1):579-597. ZHANG Xuhui,YANG Wenjuan,XUE Xusheng,et al. Challenges and developing of the intelligent remote controlon roadheaders in coal mine[J]. Journal of China Coal Society,2022,47(1):579-597.
[7] 董芳凯,赵美卿,黄伟龙. 煤矿井下暗光环境人员行为检测研究[J]. 工矿自动化,2025,51(1):21-30,144. DONG Fangkai,ZHAO Meiqing,HUANG Weilong. Research on mine worker behavior detection in low-light underground coal mine environments[J]. Journal of Mine Automation,2025,51(1):21-30,144.
[8] 骆津津,陈伟,田子建,等. 基于YOLOv8−ECW的井下人员行为实时检测算法[J]. 矿业科学学报,2025,10(2):316-327. LUO Jinjin,CHEN Wei,TIAN Zijian,et al. Real-time detection algorithm of underground personnel behavior based on YOLOv8-ECW[J]. Journal of Mining Science and Technology,2025,10(2):316-327.
[9] 韩康,李敬兆,陶荣颖. 基于改进YOLOv7和ByteTrack的煤矿关键岗位人员不安全行为识别[J]. 工矿自动化,2024,50(3):82-91. HAN Kang,LI Jingzhao,TAO Rongying. Recognition of unsafe behaviors of key position personnel in coal mines based on improved YOLOv7 and ByteTrack[J]. Journal of Mine Automation,2024,50(3):82-91.
[10] 阮进林,高鹏,孙源,等. 基于轻量级OpenPose的井下人员行为检测算法研究[J]. 煤炭工程,2024,56(4):150-156. RUAN Jinlin,GAO Peng,SUN Yuan,et al. Detection algorithm of underground personnel behavior based on lightweight OpenPose[J]. Coal Engineering,2024,56(4):150-156.
[11] SHAO Xiaoqiang,LIU Shibo,LI Xin,et al. Rep-YOLO:an efficient detection method for mine personnel[J]. Journal of Real-Time Image Processing,2024,21(2). DOI: 10.1007/S11554-023-01407-3.
[12] XIN Fangfang,HE Xinyu,YAO Chaoxiu,et al. A real-time detection for miner behavior via DYS-YOLOv8n model[J]. Journal of Real-Time Image Processing,2024,21(3). DOI: 10.1007/S11554-024-01466-0.
[13] 张帆,张嘉荣. 基于深度学习的矿井智能目标检测技术研究综述[J/OL]. 煤炭科学技术:1-14[2025-01-03]. http://kns.cnki.net/kcms/detail/11.2402.TD.20240903.1434.002.html. ZHANG Fan,ZHANG Jiarong. Research review on intelligent target detection technology for mines based on deep learning [J/OL]. Coal Science and Technology:1-14 [2025-01-03]. http://kns.cnki.net/kcms/detail/11.2402.TD.20240903.1434.002.html.
[14] 闫建星. 煤矿井下图像去雾与综采工作面异常视觉检测方法研究[D]. 西安:西安科技大学,2022. YAN Jianxing. Research on image defogging in coal mine and abnormal visual detection method in fully mechanized mining face[D]. Xi'an:Xi'an University of Science and Technology,2022.
[15] XIONG Yuwen,LI Zhiqi,CHEN Yuntao,et al. Efficient deformable convnets:Rethinking dynamic and sparse operator for vision applications[C]. The IEEE/CVF Conference on Computer Vision and Pattern Recognition,Seattle,2024:5652-5661.
[16] CHEN Jierun,KAO S H,HE Hao,et al. Run,don't walk:chasing higher FLOPS for faster neural networks[C]. The IEEE/CVF Conference on Computer Vision and Pattern Recognition,Seattle,2023:12021-12031.
[17] MA Xiaolong,GUO Fuming,NIU Wei,et al. PCONV:the missing but desirable sparsity in DNN weight pruning for real-time execution on mobile devices[J]. Proceedings of the AAAI Conference on Artificial Intelligence,2020,34(4):5117-5124. DOI: 10.1609/aaai.v34i04.5954
[18] WAN Dahang,LU Rongsheng,SHEN Siyuan,et al. Mixed local channel attention for object detection[J]. Engineering Applications of Artificial Intelligence,2023,123. DOI: 10.1016/j.engappai.2023.106442.
[19] SU Qinghua,MU Jianhong. Complex scene occluded object detection with fusion of mixed local channel attention and multi-detection layer anchor-free optimization[J]. Automation,2024,5(2):176-189. DOI: 10.3390/automation5020011
[20] ZHANG Xin,LIU Chen,YANG Degang,et al. RFAConv:innovating spatial attention and standard convolutional operation[EB/OL]. [2025-01-03]. https://arxiv.org/abs/2304.03198v6.
[21] LING Zhipeng,XIN Qi,LIN Yiyu,et al. Optimization of autonomous driving image detection based on RFAConv and triplet attention[EB/OL]. [2025-01-03]. https://arxiv.org/abs/2407.09530v1.
[22] YANG Wenjuan,ZHANG Xuhui,MA Bing,et al. An open dataset for intelligent recognition and classification of abnormal condition in longwall mining[J]. Scientific Data,2023,10(1). DOI: 10.1038/s41597-023-02322-9.
[23] JOSHI R B D,JOSHI D. MoveNet:a deep neural network for joint profile prediction across variable walking speeds and slopes[J]. IEEE Transactions on Instrumentation Measurement,2021,70. DOI: 10.1109/tim.2021.3073720.
[24] KHANAM R,HUSSAIN M. YOLOv11:an overview of the key architectural enhancements[EB/OL]. [2025-01-03]. https://arxiv.org/abs/2410.17725v1.
计量
- 文章访问数: 96
- HTML全文浏览量: 12
- PDF下载量: 18
